Releases: glotzerlab/hoomd-blue
Releases · glotzerlab/hoomd-blue
v6.1.1
Fixed
v6.1.0
v6.0.0
Added
- Implemented rigid body compatibility with MPCD without MPI (#2136).
- Add thermodynamically consistent frictional contact forces:
hoomd.md.pair.friction(#2116). create_bodiesmethod takes optionalmassesargument to set masses (#2169).
Fixed
- Use the provided alpha parameter in
make_pppm_coulomb_forces(#2153). - Add a unit test to verify that the export name of
hoomd.hpmc.compute.FreeVolume, and resolved the existing export name conflicts (#2163). - Scale move sizes correctly when the acceptance rate is 0 (#2174).
Changed
- Setting two tuples of type parameters like this
lj.r_cut[('A', 'B'), ('C', 'D')] = ...now sets the parameters A-B and C-D. This is consistent with the behavior when setting more than two parameters. For example:lj.r_cut[('A', 'B'), ('C', 'D'), ('E', 'F')] = ...sets parameters for A-B, C-D, and E-F. In previous HOOMD-blue releases, the two-tuple code path was different (it would set A-C, A-D, B-C, and B-D). Update your scripts accordingly. (#2157)
v5.4.0
v5.3.1
Fixed
- Ensure that GPU devices have concurrent unified memory capabilities (#2099).
- Fix segfault when attaching nlists with meshes (#2089).
- Install cuh headers (#2091).
- Reduce the time needed to generate mesh bonds (#2097).
- Equations displayed in
hoomd.hpmc.compute.SDFdocumentation (#2096). - Support CUDA 12.9 (#2102).
v5.3.0
Added
- The Zetterling pair potential:
hoomd.hpmc.pair.Zetterling(#2057). hoomd.write.GSDnow automatically flushes on frame writes 10 seconds or longer since the previous flush. Configure this time withauto_flush_period. The flush does not occur on a timer -- it is only called after a normally scheduled frame write (#2085).
Changed
- The
GSDwrite buffer size now defaults to 1 MiB. Files will thus grow in size more continuiously (the previous default was 64 MiB). File size changes are subject to additional buffering by the OS and may or may not be predictable. At the same time, new frames will no longer be available for reading until after the file is flushed (which occurs on write after 10 seconds by default) or closed (#2085).
Fixed
- The formulas in the docs for
variant.box.Interpolateare now consistent (#2060). - Do not increment z image in 2D simulation boxes (#2071).
- Code block example in
hoomd.md.pair.DPDConservative(#2084). - Equations displayed in
hoomd.md.methods.thermostats.Bussidocumentation (#2081). - Do not issue "Cannot acquire access to array in use" error when using FIRE on the GPU with
integrate_rotational_dof=True(#2082).
v5.2.0
Added
- Set constituent particle velocities for rigid bodies (#2024).
- Mesh tutorial (hoomd-examples#160).
Changed
- Use KaTeX to render math equations in the documentation (#2053).
- The "Organizing and Executing Simulations" tutorial now demonstrates the use of row (hoomd-examples#155).
Fixed
v5.1.1
Fixed
- Prevent warnings about forces that provide torques (or not) (#2015).
- Compile without errors or warnings with CUDA 12.8 (#2019).
force.Activeapplies torques correctly when the filter is notAll(#2020).- MD integrators no longer integrate the z degree of freedom in 2D simulation boxes (#2021).
- HPMC integrators no longer miss pairwise interactions when updaters, such as
RemoveDriftmove particles (#2022).